swin-tiny-patch4-window7-224-finetuned-fish
This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.4264
- Accuracy: 0.9412
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 75
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| No log | 0.8 | 1 | 1.8035 | 0.2941 |
| No log | 1.6 | 2 | 1.7861 | 0.2941 |
| No log | 2.4 | 3 | 1.7554 | 0.2941 |
| No log | 4.0 | 5 | 1.6954 | 0.3529 |
| No log | 4.8 | 6 | 1.6780 | 0.4118 |
| No log | 5.6 | 7 | 1.6536 | 0.4118 |
| No log | 6.4 | 8 | 1.6222 | 0.4118 |
| 1.6467 | 8.0 | 10 | 1.4682 | 0.5294 |
| 1.6467 | 8.8 | 11 | 1.3261 | 0.5294 |
| 1.6467 | 9.6 | 12 | 1.1888 | 0.5294 |
| 1.6467 | 10.4 | 13 | 1.0433 | 0.5294 |
| 1.6467 | 12.0 | 15 | 0.8212 | 0.5882 |
| 1.6467 | 12.8 | 16 | 0.7240 | 0.7059 |
| 1.6467 | 13.6 | 17 | 0.6390 | 0.8235 |
| 1.6467 | 14.4 | 18 | 0.5594 | 0.8824 |
| 0.782 | 16.0 | 20 | 0.4647 | 0.8235 |
| 0.782 | 16.8 | 21 | 0.4264 | 0.9412 |
| 0.782 | 17.6 | 22 | 0.3983 | 0.9412 |
| 0.782 | 18.4 | 23 | 0.3760 | 0.9412 |
| 0.782 | 20.0 | 25 | 0.3751 | 0.8824 |
| 0.782 | 20.8 | 26 | 0.3553 | 0.8824 |
| 0.782 | 21.6 | 27 | 0.3161 | 0.8824 |
| 0.782 | 22.4 | 28 | 0.2706 | 0.9412 |
| 0.3228 | 24.0 | 30 | 0.2100 | 0.9412 |
| 0.3228 | 24.8 | 31 | 0.1885 | 0.9412 |
| 0.3228 | 25.6 | 32 | 0.1727 | 0.9412 |
| 0.3228 | 26.4 | 33 | 0.1818 | 0.9412 |
| 0.3228 | 28.0 | 35 | 0.1959 | 0.8824 |
| 0.3228 | 28.8 | 36 | 0.1889 | 0.9412 |
| 0.3228 | 29.6 | 37 | 0.1995 | 0.8824 |
| 0.3228 | 30.4 | 38 | 0.2093 | 0.8824 |
| 0.2375 | 32.0 | 40 | 0.1869 | 0.9412 |
| 0.2375 | 32.8 | 41 | 0.1648 | 0.9412 |
| 0.2375 | 33.6 | 42 | 0.1576 | 0.9412 |
| 0.2375 | 34.4 | 43 | 0.1709 | 0.9412 |
| 0.2375 | 36.0 | 45 | 0.1717 | 0.9412 |
| 0.2375 | 36.8 | 46 | 0.1783 | 0.9412 |
| 0.2375 | 37.6 | 47 | 0.1993 | 0.8824 |
| 0.2375 | 38.4 | 48 | 0.2085 | 0.8824 |
| 0.1897 | 40.0 | 50 | 0.2028 | 0.8824 |
| 0.1897 | 40.8 | 51 | 0.1704 | 0.9412 |
| 0.1897 | 41.6 | 52 | 0.1520 | 0.9412 |
| 0.1897 | 42.4 | 53 | 0.1325 | 0.9412 |
| 0.1897 | 44.0 | 55 | 0.1451 | 0.9412 |
| 0.1897 | 44.8 | 56 | 0.1664 | 0.9412 |
| 0.1897 | 45.6 | 57 | 0.1927 | 0.8824 |
| 0.1897 | 46.4 | 58 | 0.2202 | 0.8824 |
| 0.1676 | 48.0 | 60 | 0.2569 | 0.8824 |
| 0.1676 | 48.8 | 61 | 0.2748 | 0.8824 |
| 0.1676 | 49.6 | 62 | 0.2612 | 0.8824 |
| 0.1676 | 50.4 | 63 | 0.2414 | 0.8824 |
| 0.1676 | 52.0 | 65 | 0.1842 | 0.8824 |
| 0.1676 | 52.8 | 66 | 0.1597 | 0.8824 |
| 0.1676 | 53.6 | 67 | 0.1447 | 0.8824 |
| 0.1676 | 54.4 | 68 | 0.1359 | 0.9412 |
| 0.1452 | 56.0 | 70 | 0.1367 | 0.9412 |
| 0.1452 | 56.8 | 71 | 0.1402 | 0.9412 |
| 0.1452 | 57.6 | 72 | 0.1462 | 0.8824 |
| 0.1452 | 58.4 | 73 | 0.1515 | 0.8824 |
| 0.1452 | 60.0 | 75 | 0.1585 | 0.8824 |
Framework versions
- Transformers 4.40.0.dev0
- Pytorch 2.2.1+cu121
- Datasets 2.19.1
- Tokenizers 0.15.0